Ai generated photos / art

aKilter

All-Pro
Name
David
Ai is having an increasing influence in photography. Certainly in our phones, but also in editing software and, more recently, in cameras themselves.

With this in mind, I thought it might be interesting to have a thread composed of images generated entirely by ai. This is a trend (like it, or not) that is rapidly evolving and having an enormous impact in the world of art, including photography.

If you haven’t yet had any experience with ai generated images, a quick search will provide an abundance of apps and programs for you to try. However, be forewarned…..it can be very addictive!

Here’s a few portraits to get started…..
3C36F628-6087-46FD-ADEA-55E8F3F14806.jpeg
E4B4ABE4-C96D-419F-A337-54E0C0E52DC8.jpeg
3CC2E0F3-FC05-464C-B120-1B90D5151C4B.jpeg
6D3F9FE0-753B-44F8-91F6-EFE933E8F878.jpeg
D82EA32C-A6A3-4179-8767-114AEC28F1C8.jpeg
0D6DEDFC-F845-43B1-A1C9-AABEC5D47B32.jpeg
 
Last edited:
I was playing around with these AI things over the Christmas break. As you say, it's quite amazing what can be done. Throw in the apps that can generate a landscape given a GPS position and direction of view and we're reaching the point now where getting out and taking a photo with a camera is more of an exercise in something to "do" rather than something to "create".
 
I find AI generated imagery of a different value for me personally. My mind has always haunted me with vivid imagination from dreams to ideas and daydreams my entire life. I always struggled to put them in expression, I tried poetry when I was a kid, then I switched to drawing in my mid to late teens and the final step was photography.
I haven't done drawing in some years now but I wanted to go back but it's harder now to make my hand make what my mind thinks ... Well until AI Generated Imagery, I tried Midjourney a couple of weeks ago (free 25 results but for some reason I managed to get about 38).

I went with something that has been on my mind ever since I discovered Hallow Coves band and it's become an addiction as it creeps into my dreams at night and in my daydream for the last 12 months:

W0lfy_father_walking_with_daughter_and_small_dog_up_a_hill_to_a_f490fee3-ed78-4117-a5d5-df7fd6...png


W0lfy_father_walking_with_daughter_and_small_dog_up_a_hill_to_a_52c55f79-7bac-4c99-93e5-6157f0...png

After losing my unborn daughter there's little else my mind lets me feel and think except the things I always wanted to do with her.

W0lfy_father_and_daughter_sitting_on_a_tree_branch_attached_to__fe349168-2883-4a84-860c-123880...png



Yus, it's very addictive to ask the AI to create something that is locked inside the mind. I want so many of the things I wished for to be generated by the AI, memories that are not real in what is real but what is real for what I want more than anything ... Quite addictive.
 
Interesting, which AI tools are useful to generate pictures, text etc.
Please add url link(s), thanks.
Midjourney https://midjourney.com Midjourney

DeepAI https://deepai.org › machine-learnin... Text To Image - AI Image Generator API

OpenAI https://openai.com › blog › dall-e DALL·E: Creating Images from Text

There are many more as they are popping out like mushrooms after a rainy day, some are free but with limitations) and some you have to pay for. Depending on your level of skill you might want to look into the text to generation ones (more complex ones you can train the AI on your own data set of images).
 
Most of them are just different UI interface to interact with the generator, all of them use the same AI dataset and methodologies.

Funny thing is OpenAI warned everyone not to make products out of it because it's a 2nd generation prototype for testing only and it has no underlying financial system so sustain it, the program itself will be terminated and replaced with future revisions. But a lot of people are trying to make a quick buck as fast as possible, especially the developers for mobile apps.
 
Interesting, which AI tools are useful to generate pictures, text etc.
Please add url link(s), thanks.
In my case the model is trained via dreambooth
You can do it the hard way on your (or rented) hardware, but there are services you can use to train a model in your pictures on the cheap.
astria.ai was quite good and gave downloadable models like the ones you can build yourself - so you can save the thinking on choosing the pictures from where to start

When you have a trained model you can use stable diffusion to produce images from a prompt.
I did it with an app called draw things which runs on newer macs and iPad/iPhones (you need recent ones like m1/m2 macs), otherwise there are services to do it which charge per query or per image

As my model is trained on images of a real person I like to keep the image generation at home and used both an iPhone and a m1 Mac mini
 
Most of them are just different UI interface to interact with the generator, all of them use the same AI dataset and methodologies.

Funny thing is OpenAI warned everyone not to make products out of it because it's a 2nd generation prototype for testing only and it has no underlying financial system so sustain it, the program itself will be terminated and replaced with future revisions. But a lot of people are trying to make a quick buck as fast as possible, especially the developers for mobile apps.
I asked ChatGPT to respond to your message. This was how it answered-

That is correct. OpenAI initially released GPT-3 as a research tool for developers and researchers to experiment with and test its capabilities. However, despite the warning from OpenAI, some individuals and companies have tried to commercialize the technology by creating various products and applications. This is in spite of the fact that GPT-3 is not a finished product and may be subject to changes or termination in the future.

This is the first time I’ve used it. I’m impressed enough, that I think I will try this on a work email 😂
 
Last edited:
I asked ChatGPT to respond to your message. This was how it answered-

That is correct. OpenAI initially released GPT-3 as a research tool for developers and researchers to experiment with and test its capabilities. However, despite the warning from OpenAI, some individuals and companies have tried to commercialize the technology by creating various products and applications. This is in spite of the fact that GPT-3 is not a finished product and may be subject to changes or termination in the future.

This is the first time I’ve used it. I’m impressed enough, that I think I will try this on a work email 😂
This is probably the first time on my existence that I was agreed with on something :p ... Please let AI become true if this keeps up, I need someone to agree with me more often :p

Mind you ChatGPT-3 has all the internet data up to early 2022 I believe so it doesn't have the most current information to draw upon. Also when it's wrong about something it's very confidently in being wrong, always subject it to a preview/proofing before you sent that final email to avoid problems down the road.
 
Which service did you use for this? I’ve been playing with Dalle, and can’t get a photorealistic human face to save my life. They all look so fake.

This is the monstrosity I created. It’s some real art 🙄
View attachment 362974
Try full size portrait prompts instead of wide angles. I think the AI is confused on what a human face or extremeties look like. I have notice that as well ... It turned my dog into a tiny lama:
W0lfy_father_walking_with_daughter_and_small_dog_up_a_hill_to_a_52c55f79-7bac-4c99-93e5-6157f0...png

Cute but it doesn't wip the lama's ass (Winamp reference right there)
 
This is probably the first time on my existence that I was agreed with on something :p ... Please let AI become true if this keeps up, I need someone to agree with me more often :p

Mind you ChatGPT-3 has all the internet data up to early 2022 I believe so it doesn't have the most current information to draw upon. Also when it's wrong about something it's very confidently in being wrong, always subject it to a preview/proofing before you sent that final email to avoid problems down the road.
Meh, sounds like me. When I’m wrong, I’m also very confidently wrong. No one will notice a robot is doing my emails 😂
 
Back
Top